6 Are the Androids Dreaming Yet? king, gets up, and leaves the room. Kasparov has resigned, Deep Blue has beaten him 3% to 2% points and is now the most powerful chess player on the planet. Later, when interviewed about his experience, Kasparov thought Deep Blue must have been assisted by humans during the games because the program appeared to play intuitively. The rules of the tournament allowed humans to work on the program between matches, but not during actual play. The argument has never been settled, and Deep Blue was long ago dismantled. These days chess players avoid big public matches against computers, arguing it is really a different sort of game. A computer’s ability to crunch mathematically through all the many possibilities means a chess player must play without error against a machine, but can play a more interesting and fluid match against a fellow human. Chess is computer-friendly because it is a finite problem. You always win, lose or draw. The game can't go on forever because any position that repeats itself more than three times is declared a draw, and if a player makes 50 moves without moving a pawn or taking a piece, the game is also declared a draw. In a typical game, each player makes 40 moves, and on each turn you can choose from 30 possible moves. Although this equates to a huge number of options, it is still a finite number. It is possible, therefore, to create a perfect chess-playing machine. Such a machine would project any position it encountered through every permutation to the endgame. But, although chess is solvable using brute force this might not be practical in our Universe. The storage required to hold all the possible positions being analyzed would be vast — needing most of the atoms in the Universe. You would need to pack this information into a small enough space to allow fast retrieval in order to play the first 40 moves in two hours. This would require storing all the information within a sphere no larger than three ligh